[port-peer-review] reviews
Here are my reviews of the articles other than mine.
A bit earlier than I expected and more provocative too. Well,
at least this is not anonymous and the authors can answer
(and they still have my own article to review :-)). (01)
Cheers, (02)
Philippe (03)
P.S. Eugene, if that helps you, the reviews below are also in
http://meganesia.int.gu.edu.au/~phmartin/WebKB/articles/iccs02/reviewsForPORT.html (04)
Review 1
-------- (05)
Paper's title: Creating Conceptual Access:
Faceted Knowledge Organization in the Unrev-II email archives (06)
Paper's author: Kathryn La Barre and Chris Dent (07)
Summary. The authors describe an experiment to use, combine and
compare various document indexation tools (especially in latent semantic analysis)
for the creation of "conceptual" clusters of terms/concepts/facets/documents
(however, the only relations between the terms/concepts/facets/documents/clusters
seem to be measures of similarity/divergence calculated by the semantic analysis
tools). The evaluation (and refinement) of the created clusters, and hence what
the authors call the "access structure" of the set of documents, is expected to be
done by people via a manual ranking and tagging (keywords or short phrase) of
documents. It is also expected that each phase of refinement will be usable as
input for another phase of automated clustering. (08)
Clarity and precision. I found the description too general and hence difficult
to understand. There are many repetitions but at the same level of generality,
i.e. without precision, definition or example, even about the most frequently
used words/expressions: "cluster", "facet", "access structure", "coding messages".
The figures are not much helpful since they do not show any term/concept/facet and
the nature of the relations between the nodes is not explicited.
My understanding of the article relies on the information that the output of
classic document indexation tools is used, and my conviction that therefore
there cannot be anything really "conceptual" or "structured" to exploit.
Hence, "facet" must refer to a simple keyword, and "access structure" to some
calculated similarity relations which do not have any commonsense meaning.
Which document indexation tools have been used, and how, is also unspecified. (09)
Originality. The proceedings of the WWW conferences are full of descriptions
of tools creating clusters of documents (based on classic document indexation
techniques) and permitting to navigate within and between them. I do not know
(or like the output of) these approaches enough to appreciate their originality.
I was more surprized by the absence of references to the use of
Formal Concept Analysis (or similar methods) for the structuration and navigation
of a base of documents and the terms used for indexing it. Indeed, this approach
has the advantages of producing a genuinely understandable and structured index
on the documents. It is also quite common now. In the CG community,
(i) Guy Mineau classified documents in that way, and permitted navigation in the
lattice, about 10 years ago now;
(ii) this is of course a classic application for the Darmstadt group;
(iii) in my own team (KVO, http://www.kvocentral.com/), Richard Cole has
finished a PhD thesis, partially about this subject too, about a year ago.
One of his tools is specialized for e-mails classification/access and is called
"Email Concept Analysis" tool. Navigation along specialization links
between concepts (term/document sets), filtering according to attributes
(term, email author, destinations, date, subject, ...) and
generation of FCA contexts/scales is possible. I mention it as a comparison to
the navigation envisaged by the authors of the article. (010)
Interest of the approach. This question is not discussed in the article.
The abstract mentions that the hypothesis is that classic document indexation
techniques "may be worthwhile tools to generate meaningful clusters in the
dataset". However, there is no indication this is so in the article.
I personally do not think this is so. I do not even have much interest in using
the "Email Concept Analysis" tool of my friend because it is not enough
"conceptual"/"knowledge-based" to me: I am not interested in retrieving sets
of e-mails/documents according to terms/authors/..., I am interested in getting
precise answers to precise questions, in seeing all the
(conceptual/rhetorical/argumentation) relations from one object (category or
sentence) and navigate along the relations (e.g. navigate along an argumentation
path). Classic or FCA-based "document-indexation" techniques cannot provide that.
Part of the problem is that each document includes many sentences (facts, ideas,
argumentation for ideas, etc.) and sentences related to an idea/object
are scattered in many documents. Only knowledge-based techniques permit
to collect (and organize, if necessary) the relevant sentences to answer
precisely to a precise query. Of course, until natural languages parsing
techniques actually permit to extract the meaning of sentences in documents,
the downside of the approach is that knowledge bases have to be built
more or less manually. Certain uses of informal methods like Topics Maps might
offer an acceptable compromise between precision and ease-of-representation. (011)
Review 2
-------- (012)
Paper's title: Making Doug's Dream Come True: Collaboratories in Context (013)
Paper's author: Aldo De Moor (014)
Summary. The author claims that a high-level design approach (or "framework",
or set of "design principles") such as Douglas Engelbart's is needed to design
and integrate collaborative tools. Some distinctions (design principles?)
are presented, and then used to present a few requirements for PORT. (015)
Clarity and precision. I do not know any precise, clear and easily appliable
framework/methodology, e.g. in software engineering, knowledge engineering, and
that includes the one I developped for "explanation generation" (as an extension
and application of KADS) during my Honours thesis. I do not think there is an
escape to that. (016)
Originality. Other frameworks are not cited although I guess this one
competes or is complementary to quite a few. (017)
Interest of the approach. I am not convinced the few very general
distinctions presented in the article are really guiding. Was the author
guided by them to come up with the content (not the presentation) of his
few notes on PORT? Wasn't this content already common knowledge and only
a glimpse of the tip of the iceberg of the actually needed requirements?
Ok, these notes are examples only but important requirements such as
lexical/structural/semantic/ontological conventions (e.g. such as those
I presented at ICCS'00) are not mentionned, nor is the need for
knowledge-based cooperation protocols (asynchronous such as those in
WebKB-2 or CO4, or synchronous shuch as those in Tadzebao and
WebOnto).
Can any high-level methodology or set of "design principles" for
"collaboratories" lead developpers to come up with new ideas, or ease the
technical integration of already developped tools? (018)
Miscellaneous. Assuming that "links" do not impose constraints but only
bring more information that may be exploited if necessary, I do not understand
the last part of the second sentence in the following quote:
"The more system processes are linked, the higher the level of integration.
In some cases, a high level of integration may be desirable, whereas in other
cases, having loosely coupled processes may be more useful."
http://lab.bootstrap.org/port/papers/2002/demoor.html#nid021 (019)
Review 3
-------- (020)
Paper's title: On the practical bearings of Peirce's maxim (021)
Paper's author: Janos J. Sarbo (022)
Summary. The author presents certains distinctions among signs and,
if I understood, claims that thanks to these distinctions, the signs
can be recursively and automatically combined to represent knowledge,
and for people, to understand/interpret the real world. (023)
Clarity and precision. At a sufficiently high level of abstraction, guessing
and describing what might happen in people's brains is not a particularly risky
or difficult activity. There is an infinity of possible abstractions and
any high level reasonable model is unlikely to be proven wrong. But how does
that help us to implement the human functions? This article claims the
presented model is a computational model and that the sign can be "generated"
via their automatic combination. How? And how are the presented distinctions
a guide? Furthermore, what is the expected input: formal representations
already taking into account the proposed distinctions, free natural language,
real world (audio/visual sensors)? In the last two cases, the additional problem
is of course to reach the first case. (024)
Originality. I do not have enough information to comment on that. (025)
Interest of the approach. I do not have enough information to comment on that. (026)
Miscellaneous.
In http://lab.bootstrap.org/port/papers/2002/sarbo.html#nid08, the author
says an ideal language would allow everything to be viewed as an argument
and permit everything to be recognized as an argument. Well, LISP permits
about everything to be used as an argument, even code itself, and what
developpers may be able to program to "recognize" arguments seems
more limited by their time and abilities than the language itself. What more
abstract representational function is not and cannot be covered using LISP-like
syntaxes? I am likely to be on the wrong track.
In http://lab.bootstrap.org/port/papers/2002/sarbo.html#nid09, the author
argues that human KR is based on signs. What's the alternative? If symbols
are signs, every language (every symbolic representation means) is based
on signs. (027)